Oxygen enters the body with inspiration and diffuses from the lungs into the blood. Subsequently the blood circulation transports oxygen to the tissues. Disorders of oxygen transport from the inspired air into the blood can result in a low oxygen saturation of the blood. These disorders in oxygen uptake include abnormal ventilation of the lung, seen in for example chronic obstructive pulmonary disease; abnormal oxygen diffusion in the lung, seen in for example pulmonary fibrosis; and abnormal perfusion (i.e. blood flow) through the lung. Estimation of parameters describing these oxygenation problems is important for diagnosis, monitoring and assessing appropriate therapeutic intervention. This is true in a wide variety of patients, from those who are automatically ventilated and who often require continuous supplement of oxygen, to out-patients who only suffer from dyspnoe during exercise.
In clinical practice the clinician usually relies upon simple measurements or variable estimates to assess the patients oxygenation problems. These include qualitative estimates obtained from stethoscopy or chest X-ray. They also include more quantitative estimates such as arterial oxygen saturation, the alveolar-arterial oxygen pressure gradient, or estimates of the “effective shunt”, a parameter which describes all oxygenation problems in terms of a fraction of blood which does not flow through the lungs (Siggaard-Andersen and Siggaard-Andersen, 1985).
Whilst the “effective shunt” is a parameter which has been used widely in the clinical literature it cannot adequately describe the ‘clinical’ picture seen in patients when the inspired oxygen fraction is varied. This observation is illustrated in FIG. 1 where the “effective shunt” has been estimated for a single patient at four different inspired oxygen fractions, and varies from 15–25%.
In contrast to the poor clinical description of oxygenation problems, detailed experimental techniques such as the Multiple Inert Gas Elimination Technique (MIGET) (Wagner et al., 1974) have been developed which describe the parameters of models with as many as fifty lung compartments. The parameters of these models give an accurate physiological picture of the patient. Whilst the MIGET has found widespread application as an experimental tool its use as a routine clinical tool has been somewhat limited (Wagner et al., 1987). This is largely due to the cost and complexity of the technique.
As stated previously, “effective shunt” is insufficient to describe oxygenation problems. Further parameters describing the patient's oxygenation problem can be obtained from data where inspired oxygen is varied, i.e. data similar to that presented in FIG. 1. This was first recognised by Riley et al. (1951a, 1951b) and later by King et al. (1974). These authors used mathematical models to divide the oxygenation problem into that due to an alveolar-lung capillary drop in the partial pressure of oxygen, and that due to a shunt problem. To estimate two parameters describing the oxygenation problem requires taking measurements of blood samples and of ventilatory variables at each inspired oxygen fraction. Estimating lung parameters using the data from four inspired oxygen fractions required four blood samples, a procedure which is still rather time consuming and in some environments impractical.
More recently, development of non-invasive methods for measuring the oxygen saturation of the blood have lead to renewed interest in estimation of parameters describing oxygen transport obtained by varying FIO2. Andreassen et al. (1996, 1999), Sapsford et al. (1995), de Gray et al. (1997) and Roe et al. (1997), have presented the use of two parameter mathematical models of oxygen transport, the oxygenation problem being described as shunt combined with either a diffusion abnormality (Andreassen et al. (1996, 1999)) or due to a ventilation/perfusion () mismatch (Sapsford et al. (1995), de Gray et al (1997), Roe et al., (1997)). These model representations have been shown to provide identical fits to routine blood gas and ventilatory data obtained by varying FIO2 (Rees et al. 1997).
The clinical relevance of the two parameter models is illustrated in FIG. 2, where increases in the pulmonary shunt parameter results in a vertical depression of the FIO2/SaO2 curve, (V-shift) and abnormalities in the second parameter (ventilation/perfusion () mismatch or oxygen diffusion resistance (Rdiff)) results in a lateral displacement of the FIO2/SaO2 curve. Clearly, the lateral displacement of the FIO2/SaO2 curve (H-shift) is clinically a more significant problem as it describes a situation where large changes in oxygen saturation can occur for only small changes in FIO2. In this situation the patient is at increased risk of an oxygenation problem.
The two parameter model of Sapsford et al. (1995), has been shown to fit data from normal subjects; patients before and after thoracotomy (Sapsford et al. 1995, de Gray et al., 1997); and patients during (Sapsford et al. 1995, Roe et al., 1997), and after (Roe et al., 1997) abdominal surgery. Similarly, the two-parameter model described by Andreassen et. al. has been shown to fit data from normal subject and postoperative cardiac patients (Andreassen, 1999) and a wide range of as yet un-published results. Examples of these results are shown in FIG. 3.
In contrary to detailed experimental approaches (e.g. the MIGET), these two parameter models can be used routinely in clinical practice. In particular, these techniques may find application in the monitoring and choice of therapeutic treatment for patients with left-sided heart failure, or to assess patients risk of post-operative hypoxaemia.
Until now, estimation of oxygenation parameters has involved manual titration of the FIO2/SaO2 curve and off-line estimation of the parameter values. This is time consuming with experimental times of approximately 45 minutes, not including the time required for off line parameter estimation. This limits the use of the method as a clinical tool.